The focus of this study is to improve walking symmetry in order to optimize walking ability and reduce disability for Veterans with non-traumatic lower limb amputation (LLA). Over 80% of current LLAs are non-traumatic, resulting from complications of pathologies, such as diabetes mellitus and peripheral artery disease. Despite current declines in total amputation rate among Veterans, the population with non-traumatic LLA is growing. For example, from 2000 to 2004 the relative amputation rate decreased by 34%. However, due to an increase in the number of Veterans with diabetes during the same period, the population of Veterans with diabetes and initial LLA increased by 23%. Following LLA, excessive gait asymmetry is common. Asymmetric gait characteristics are of critical importance as excessive asymmetry may increase the severity of disability experienced by people with non-traumatic LLA and contribute to secondary pain conditions (low back pain and osteoarthritis), poor gait efficiency, declines in physical performance, and compromised skin integrity of the residual limb. Compared to individuals with traumatic LLA, the poor gait performance of those with non- traumatic LLA is further compounded by older age, lower premorbid function, presence of comorbidities, frequent wound development, and delayed healing in the residual limb. While improving gait symmetry is a goal of conventional prosthetic rehabilitation, persistence of gait asymmetry for years after LLA highlights the ineffectiveness of current rehabilitation practices in achieving this goal. As a means of improving gait symmetry, this study aims to determine the efficacy of error-manipulation gait training using two approaches (error-augmentation and error-correction) compared to current standard-of-care in a three-arm randomized controlled trial. Error-manipulation gait training intervention will be delivered in eight training sessions (2x/week, 4 weeks) with 54 Veterans (18 per intervention group, 18 in control group) who have non-traumatic, unilateral, transtibial LLA. It is unclear which, if either, form of error-manipulation gait training is efficacious for improving persistent gait symmetry in Veterans with non-traumatic LLA. Error-augmentation gait training is a promising and novel intervention that involves exaggerating an existing movement error to force the neuromuscular system to correct the error. While this form of gait training improved gait symmetry in small studies of individuals with chronic stroke or traumatic amputation, it has yet to be evaluated in older Veterans with non- traumatic LLA. In contrast, error-correction training involves reducing movement errors by overcorrecting for asymmetry. Both error-augmentation and error-correction gait training are based upon motor learning principles of distributed practice, task specificity, and feedback. Each of these error-manipulation interventions have potential advantages over traditional gait training following LLA which involves repeated bouts of walking with minimal feedback on movement quality and is often unsupervised. Therefore, the primary aim of this study is to determine the efficacy of error-manipulation gait training to improve gait symmetry. A secondary aim is to evaluate signals of efficacy for improved secondary measures of physical function. Lastly, this study will explore changes to residual limb skin health and prosthesis socket fit following error-manipulation gait training. The unique use of motor learning principles in error-manipulation gait training to improve gait symmetry addresses the problem of chronic gait asymmetry following non-traumatic LLA. The results of this study will advance rehabilitation knowledge and provide necessary evidence for the clinical translation of gait training protocols based in motor learning principles for the at-risk population of Veterans with non-traumatic LLA.